Influence of Different Hydrocarbons on Chemical Vapor Deposition Growth and Surface Morphological Defects in 4H‐SiC Epitaxial Layers

Controlled epitaxial growth of 4H‐SiC is essential for advancing both power electronics and quantum technologies. This study explores how different carbon sources—methane and propane—affect the surface morphology of these epitaxial layers. By varying C/Si ratios and using the two mentioned hydrocarbons as the carbon source in chloride‐based epitaxial growth of 4H‐SiC layers, it is unveiled that methane results in an exceptionally smooth surface. However, it pronounces surface irregularities such as short step bunching and dislocation‐related etch pits. Moreover, methane amplifies the overgrowth of triangular defects with the 4H polytype. In contrast, the introduction of propane causes a step‐bunched surface together with inclined line‐like surface morphological defects. Notably, a majority of the triangular defects exhibit a pure 3C character without an overgrown 4H polytype. It is shown that these outcomes could be attributed to different sticking coefficients and diffusivity of the molecular species resulting from different carbon sources on the 4H‐SiC surface during the epitaxial growth. This research also uncovers the underlying origins and mechanisms responsible for various surface morphological defects.


Introduction
4H-SiC stands out as a wide bandgap semiconductor material with remarkable qualities, such as high critical electric breakdown field, thermal conductivity, and saturation drift velocity.These properties make 4H-SiC highly attractive for power device applications. [1]Significant improvements in wafer enlargement and crystal quality over the last 10 years have increasingly expanded the implementation of 4H-SiC into real power electronic devices. [2]] Epitaxial layers with low basal plane dislocation (BPD) density are obtained on 4°off-cut substrates, [6] and the density of extended defects (like stacking faults and triangular defects) and point defects (like foreign impurities and vacancies) is reduced with a scoped control over the growth process. [7]0] Additionally, the potential of 4H-SiC has not been limited only to power electronics but also to advanced quantum technologies.Due to its wide bandgap, weak spin-orbit coupling, and extremely dilute nuclear-spin bath in isotopically pure layers, 4H-SiC poses outstanding quantum properties at room temperature. [11,12]ifferent solid-state emitters such as silicon vacancies (V Si ), divacancies (V Si V C ), and antisite-vacancy pairs have been identified in the 4H-SiC lattice. [13]The supremacy of 4H-SiC in material properties and advanced manufacturing supply chain (commercial availability and CMOS compatibility) promises a reliable material platform alternative to diamond for the future of quantum photonics.While diamond cannot provide a sufficient amount of spinphoton interfaces, 4H-SiC offers the possibility of processing an increased number of these interfaces on an integrated photonic chip, [14] thereby enabling the realization of tangible quantum devices and networks.
The performance of 4H-SiC-based devices is heavily influenced by crystallographic defects, such as BPDs, dislocation arrays, micropipes, and stacking faults, and hence significant efforts have been made to minimize the formation of these defects.On the other hand, although surface morphological defects are not as detrimental as crystallographic defects, they are still unfavorable for both types of 4H-SiC-based electronic and photonic devices.
[17][18] This has made the growth of extremely thick epitaxial layers possible, moving a great step forward toward the realization of ultrahigh power bipolar devices. [19]Epitaxial growth is typically carried out using either standard growth chemistry or chloride-based chemistry, with propane commonly serving as the C source.22][23][24] Additionally, when it comes to thick epitaxial layers, they often exhibit a pronounced density of surface morphological defects and significant surface roughness.
We have observed that the use of different hydrocarbon sources in chloride-based chemistry has a significant influence on the surface quality and formation of surface morphological defects such as short step bunching (SSB) and inclined line-like (ILL) defects, [25] as well as the structure of extended defects like triangular defects.
In this study, we investigate and control the surface morphological defects in 4H-SiC epitaxial layers by exploring the impact of different hydrocarbon sources.Multiple series of 4H-SiC epitaxial layers were prepared under varying growth conditions and were analyzed using a range of microscopic techniques.The origin and formation mechanisms of different types of surface morphological defects are investigated.

Experimental Section
2.1.In Situ Surface Preparation and Epitaxial Growth 4°off-axis substrates are cut from a 150 mm diameter standard highly doped n-type wafer.The 25 μm thick 4H-SiC epitaxial layers of this study are grown on the substrates in a horizontal hot-wall chemical vapor deposition reactor, utilizing trichlorosilane (HCl 3 Si) as the silicon precursor and either methane (CH 4 ) or propane (C 3 H 8 ) as the carbon sources and hydrogen as the carrier gas.The impact of carbon sources is investigated by growing two main sets of samples with different carbon sources, where the C/Si ratio varies ranging from 0.9 to 1.3, while keeping other growth conditions fixed.The growth parameters are provided in Table 1.Complementary samples are also grown for both of the carbon sources whereas in these samples, only one parameter is changed at a time, while all other parameters remain fixed.
Alongside the epitaxial layers, a corresponding set of substrates, taken from the same wafer, are also examined to investigate the surface properties of the substrate before growth initiation.These substrates were subjected to in situ surface preparation under various ambient conditions at the growth temperature but were not subjected to epitaxial layer deposition.Further details regarding the substrate's surface preparation are provided in Table 2.

Characterization
The samples are analyzed using differential interference contrast (DIC) microscopy, atomic force microscopy (AFM) in tapping mode, dark-field optical microscopy (DF), transmission optical microscopy, and room temperature Raman spectroscopy with excitation wavelength at 532 nm (laser spot size on the sample around 3 μm by using 60Â air objective) and a spectrograph (Andor Kymera 328i) featuring a 600 L mm À1 diffraction grating (blazed at 500 nm).To identify any correlation between specific surface morphological defects and crystallographic defects in the substrate, one sample from each group is etched in molten KOH at 500 °C for 5 min.

In Situ Substrate Surface Preparation
Figure 1a-c displays DF optical images taken from the substrates after surface preparation under different ambient conditions, whereas the bottom images are taken with AFM in tapping mode from regions marked in the optical images.A comparison of optical images shows a clear difference in the surface morphology in terms of surface roughness, density, and dimensions of SSB-related lines.
When the substrate's surface was prepared in a hydrogen ambient, short lines resembling SSB defects (encircled with a dashed line in Figure 1a), previously reported in the literature, [26][27][28] appeared on the surface with an approximate length of 100 μm.At higher magnifications (inset in Figure 1a), the triangular shape of these SSBs became apparent, with the apex oriented toward the down-step direction.
When CH 4 was added to hydrogen during the surface preparation (Figure 1b), the length of triangular features increased to ≈750 μm, resulting in severe additional local step bunching with an average step height of ≈3 nm (line profile in the AFM image of Figure 1b).Conversely, adding C 3 H 8 to hydrogen during surface preparation (Figure 1c) did not only reduce the length of the triangles to ≈70 μm but also resulted in a smooth surface without any additional local step bunching related to SSBs.Additionally, some of the SSBs appeared to have a 2-6 nm high bump-like feature at the apex of their triangular shape (encircled in Figure 1c).These types of features were also observed on the surface of the substrate etched in hydrogen ambient as well, however with  relatively lower density.A statistical analysis of these features on samples EH and EP, with SSB lengths ranging from 70 to 100 μm, revealed that ≈10% of the SSBs on sample EH had bumps, while this value raised to ≈30% for sample EP.The surface roughness root-mean-square (RMS) value of all samples, excluding the triangular features and step bunched regions, was measured to be similar, around 0.3 nm.To avoid the impact of any preintroduced hydrocarbon on the epitaxial layers' properties, substrates for growing epitaxial layers in this study were surface prepared solely in hydrogen ambient.

CH 4 as the Carbon Source
Figure 2 presents the optical and AFM images of the epitaxial layers grown using CH 4 as the carbon source with different C/Si ratios, while the rest of the growth parameters were kept fixed.The most notable characteristic of the surface of these epitaxial layers is the presence of well-pronounced SSBs, which can be observed in all samples grown using varying C/Si ratios.Moreover, the prominence of SSBs increases with the C/Si ratio, and in the special case of the sample grown under C/Si: 1.3 the surface became heavily step bunched.Additional features observed on the surface of epitaxial layers grown with CH 4 are surface pits.Identifying these pits using DIC microscopy in epitaxial layers grown at low C/Si ratios proved to be challenging.However, it was found that these pits exhibited an increasing prominence when higher C/Si ratios were employed (see Figure 2c for a visual representation of the encircled pits).AFM images in Figure 2 reveal that all samples have a smooth surface (excluding SSB regions), with an RMS value lower than 0.2 nm, indicating that the use of CH 4 has resulted in an even smoother epitaxial layer surface than the initial roughness (0.3 nm) of the starting surface prepared in hydrogen ambient.
Triangular defects (mainly composed of 3C polytype with a 3C facet at the apex) are commonly observed in 4H-SiC epitaxial layers.The epitaxial layers grown with CH 4 show overgrowth of 3C triangular defects with 4H polytype (Figure 3) and notably majority of the defects in this particular set of samples lack the commonly observed 3C facet at their apex.This was confirmed through Raman spectroscopy performed in different regions of the defects which only revealed 4H polytype-related peaks (Figure 3b).Furthermore, the DF image (Figure 3d) confirms the overgrowth of these triangular defects with the upcoming steps, forming a perfect sandwich structure of a foreign polytype (e.g., 3C-SiC) between 4H-SiC layers.The transmission image (Figure 3c) indicates that the sandwiched 3C-SiC is too thin to discern a difference in color between the defective region and the neighboring area.
Figure 4 summarizes the influence of different C/Si ratios, growth temperature, and epitaxial layer thickness on the length of SSB-related lines.The length of the SSBs remains consistent at ≈260 μm across all C/Si ratios (Figure 4a).However, analyzing sets of complementary samples revealed that the SSB length can be influenced by the growth temperature and epitaxial l  ayer thickness.The increase in SSB length follows an exponential growth by increasing the temperature (Figure 4b).On the other hand, increasing the thickness of the epitaxial layers (Figure 4c) leads to a linear increase in the SSB length.

C 3 H 8 as the Carbon Source
Figure 5 depicts the optical and AFM images of the epitaxial layers grown using C 3 H 8 as the carbon source.AFM reveals that surface step bunching is a common feature of all samples grown with different C/Si ratios.However, SSBs formed during surface preparation are only visible on the surface of the sample grown with a low C/Si ratio of 0.9 (Figure 5a).The inset in Figure 5a presents a typical triangular defect, which is similar to the triangular defects reported to be grown under a low C/Si ratio (Si-rich environment). [29]Increasing the C/Si ratio eliminates SSBs but leads to the formation of ILL surface morphological defects (Figure 5b-e). [25]Notably, two different orientations of the ILL defects can be observed, with angles of þ80 and À80 degrees to the step-flow direction.Most importantly, no such defects are observed when growth is performed using CH 4 as a C source even with a high C/Si ratio.Furthermore, the introduction of C 3 H 8 as the carbon source resulted in fewer surface pits with less visibility.
Regarding the shape of the triangular defects, in contrast to the epitaxial layers grown using CH 4 , the majority of the triangular defects had a 3C facet at the apex, confirmed by Raman spectroscopy.It is clear from Figure 6 that the 796 cm À1 peak in Raman shift pointed by the arrow in Figure 6b is related to the transverse optical phonon mode of 3C polytype and is vanishing by going toward the downstream side of the triangular defect and completely disappearing off the triangular defect.The yellow color in the transmission image also confirms the 3C nature of the defect, with the shade also agreeing with the extension of the 3C region on the basal plane of a 4°off-cut substrate along the step-flow direction.Furthermore, the discontinuation of the surface step bunching-related lines in the 3C region, observed in the DF image, implies that no overgrowth with 4H-SiC has occurred in this region.
Figure 7 presents the influence of different growth parameters (only one parameter is changed in each set) on the length of ILL defects.The length of the inclined lines increases linearly with the C/Si ratio (Figure 7a).Results from the complementary samples showed that this length is independent of the growth temperature (Figure 7b), but increases exponentially with the epitaxial layer thickness (Figure 7c).

Molten KOH Etching
To study the influence of substrate defects such as dislocations on the surface morphological defects in epitaxial layers, three selected samples were etched in molten KOH: a bare substrate after in situ surface preparation in H ambient, an epitaxial layer grown with CH 4 , and an epitaxial layer grown with C 3 H 8 as C source.After KOH etching, each sample was mapped under an optical microscope.All kinds of dislocations are preferentially etched on the Si-face and appear with characteristic etch pits that can be easily identified.observed that the SSBs formed on the substrates, which were already present after surface preparation under H 2 ambient at the growth temperature, do not exhibit a direct correlation with any of the dislocation etch pits.This indicates that the SSBs are not associated with any crystallographic defect (such as dislocations) in the substrate.In Figure 8b, an optical image of a typical epitaxial layer grown with CH 4 containing SSBs on the surface (sample M4) is shown after being etched in molten KOH.All SSB-related lines are numbered from 1 to 6 in this image.It seems like some of the SSB-related lines (nos. 4 and 6) are associated with etch pits but not all.This indicates that SSB-related lines do not necessarily originate from dislocations in the substrates but it could be that an independently formed SSB-related line in the vicinity of a threading dislocation is pinned by dislocation.Figure .8c presents a typical image of an epitaxial layer grown with C 3 H 8 , showing ILL defects on the surface (sample P4).In this case, molten KOH etching revealed that there is an associated threading screw dislocation (TSD) etch pit for every ILL defect on the surface.Three ILL defects are indicated by corresponding numbers, and the associated TSD etch pits are highlighted with outlined hexagonal shapes.

In Situ Substrate Surface Preparation
Initially reported by Li et al., [27] the appearance of SSB-like features with elongated triangular shapes on the surface of 4H-SiC substrates after in situ surface preparation raised questions about their origin.It was suggested that the retreat of steps during surface preparation could be hindered by an obstacle, resulting in the formation of elongated triangular ridges (SSBs).Later on, it was shown that these types of features may originate from surface and subsurface damage of the substrate. [30]igure 9 presents AFM images around the apex of the two types of SSBs obtained from sample EP as an illustrative example.It is evident that the two types of SSBs possess distinct dimensions.Specifically, the SSB with a bump at the apex is wider and exhibits a greater height for the triangular ridge.This suggests that the formation of these two types of SSBs likely occurred at different stages, corresponding to the varying depth of the obstacles within the 4H-SiC substrate.
The formation of SSBs without bumps can be explained by considering that such SSBs are formed through shallow scratches or microcracks on the surface, which introduce microscale voids in the near-surface region. [31]Although they can impede the step retreat during surface preparation, they gradually diminish as etching progresses, as they do not extend too deep into the material.This is evident in Figure 9a, where the width of the triangular ridge is ≈2.5 μm, narrower than the one depicted in Figure 9b, which has a width of 3 μm.
In contrast, triangular ridges with bumps are more likely to be formed due to the impact of embedded foreign particles, originating from the slurry or polishing pads.If these particles are chemically inert compared to the crystalline 4H-SiC, they cannot be easily etched away during substrate surface preparation, or at least not at the same rate as the 4H-SiC crystal.Consequently, besides hindering step retreat, a bump-like feature remains at the apex.Figure 10 presents a schematic illustrating the effect of such subsurface damage damages and embedded foreign particles on the formation of SSBs.
A chemical-mechanical polished substrate surface (Figure 10a), upon exposure to hydrogen at high temperature, is etched and reveals the surface step structure and embedded foreign particles (Figure 10b).As the substrate has a 4°off-cut and the etching primarily occurs on the step edges, step retreat commences (Figure 10c).The foreign particle pins the steps retreat and disrupts the step structure which appears in the form of a triangle with a bump-like feature at the apex (Figure 10c).However, far enough from this point, the step continues to retreat, without being affected by the obstacle (Figure 10d).Over time, this process continues, resulting in the formation of an elongated triangular ridge (i.e., SSB) on the surface (Figure 10e).It should be noted that the steps retreating toward the obstacle accumulate due to the pinning effect and form a triangular ridge-shaped region.The steps retreating from behind the obstacle create a smooth (0001) surface, start accumulating in the base region, and together with the triangular ridge form a triangular-shaped region.It is also evident from our measurements (Figure 9a,b) that a larger obstacle leads to a larger accumulation of steps in the tail region and a large area of triangular-shaped region.
In principle, introducing either source of Si or C during surface preparation is expected to reduce the etching rate. [32,33]This phenomenon may also reduce the etching rate of embedded foreign particles when introducing a C source and hence we observe a higher number of SSBs with bump-like features.However, a relatively higher etching rate under H provides more favorable conditions to completely etch away the foreign particles.
Although CH 4 or C 3 H 8 was introduced into the chamber with the same molar flow rates, Figure 1 clearly illustrates that their effects have been significantly different.It is known that CH 4 undergoes fewer decomposition reactions to turn into lighter molecules. [34,35]Considering the very short residency time of molecular species during the process (due to low pressure and heavy flow of H 2 ), the two different hydrocarbons may provide different sets of decomposed hydrocarbons in the growth zone.When C 3 H 8 is used, larger hydrocarbons reach the surface during the process, while CH 4 increases the possibility of reaching single-carbon-containing molecules to the surface.Nevertheless, a complete understanding of the chemistry is still lacking.
Additionally, different byproducts of the different hydrocarbons upon decomposition in H 2 ambient would have different adsorption rates on the 4H-SiC surface.The adsorption rate is proportional to the sticking coefficient of a molecule and is determined from simple thermodynamics: [35] ω ¼ γ In this expression, γ, R, T, M, and [X] are the sticking coefficient, molar gas constant, temperature, molar mass, and concentration in the gas phase, respectively.
In the first step of its decomposition chain, CH 4 decomposes into CH 3 and H, whereas C 3 H 8 decomposes into C 2 H 5 and CH 3 .
Table 3 shows different sticking coefficients for these species.A comparison between the different sticking coefficients for these byproducts shows that decomposed species resulting from CH 4 have lower adsorption rates compared to those resulting from C 3 H 8 .Consequently, molecular species resulting from CH 4 have higher diffusion probability on the surface of 4H-SiC. [35]hese findings help to explain the differing impacts of CH 4 and C 3 H 8 on the surface.When CH 4 is introduced, carboncontaining molecular species diffuse readily across the surface within a given time frame compared to molecular species present when C 3 H 8 is introduced.As the etching process primarily occurs at step edges and surface kinks, the increased likelihood of carbon-containing species reaching these regions leads to a reduction in the etching rate.Referring to Figure 10b, during step retreat, carbon species accumulate around regions with higher steps, further decelerating the step retreat and resulting in the formation of elongated triangular ridges.On the other hand, when C 3 H 8 is introduced, the diffusion of carbon species is lower compared to CH 4 .As a result, the accumulation of species around the aforementioned regions is less pronounced, leading to a more even distribution of species across the surface and hence a reduced etching rate and an increased number of SSBs with bump-like features.

CH 4 as the Carbon Source During Epitaxial Growth
AFM analysis of the epitaxial layers grown with CH 4 revealed an extremely smooth surface (Figure 2) with an RMS value below 0.2 nm (except for the sample M5).However, optical images showed the presence of SSB defects with random distribution on the surface of almost all samples.Figure 11a,b present AFM images from the central region and the tail region of a  typical SSB, respectively.In contrast to the triangular ridgeshaped SSBs observed on the bare substrate after in situ surface preparation, the SSBs in epitaxial layers are composed of a terrace surrounded by elongated ridges.The typical depth of the terrace is about 7.5 nm and the height of the ridges from the surrounding surface is about up to 4.5 nm, as shown in Figure 11a.
With moving toward the tail regions of the SSB, the depth of the terrace decreases and eventually levels with the neighboring surface (Figure 11b).Furthermore, regular step bunching was also observed on this sample (Figure 11c) with a typical step height of around 4.5 nm and associated terrace with 4°off facet.An observed increased length of SSB-related line in the epitaxial layer compared to its length on the as-prepared surface of the substrate can be explained by understanding the surface step-flow mechanism during the epitaxial growth around the triangular ridge shape of SSB which is the exact opposite of what it was during surface etching.During epitaxy, the growth commences in the step-flow direction ½1120 which is opposite to step flow during etching.As the growth progresses, the upcoming steps reach the elongated ridge and continue to overgrow it but eventually get pinned at the apex of the triangle.The shape of the triangle becomes inverted with the apex now in the upstep direction which continues to pin the step flow, as shown in the schematics in Figure 12a,b.If we take the example given in Figure 11, over time, the arms of the triangle which are now in the upstep direction continue to grow in height (4.5 nm) and length, hinder the step flow, and lead to the formation of the deep terrace (7.5 nm) followed by the edge of facet (4.5 nm).With the increasing thickness of the epitaxial layer, these features get more elongated and become more and more pronounced on the surface.
While we have found a plausible explanation for the elongation and prominent formation of SSBs during epitaxial growth with CH 4 as the carbon source, the specific reason for their exclusive presence with CH 4 remains unknown.However, building upon the explanation regarding the distinct diffusion coefficients of carbon sources, we can extend this reasoning to address the current question.When CH 4 is utilized as the carbon source, the molecular species involved in growth exhibit enhanced mobility enabling them to access vacant atomic sites such as step edges and kinks.Consequently, growth occurs on nearly all available sites, leading to the replication and expansion of surface features established or modified during the surface preparation step.This explanation finds support in the behavior of SSB length variation under different growth conditions, as depicted in Figure 4. Changes in the C/Si ratio do not influence the length of SSBs.This relationship is illustrated in Figure 12, where the extension of SSBs beyond their initial length is directly tied to the advancement of steps.Within a defined time frame, this progress can be influenced by the growth rate.Given that the growth is already occurring in the Si-deficient regime (high C/Si), fluctuations in the C/Si ratio do not affect the length of SSBs.Conversely, increases in growth temperature and epitaxial  layer thickness result in elongated SSBs.The elevated temperature contributes to higher etching rates, leading to more extensive retreatment of steps during surface preparation, hence a more pronounced and elongated SSB on the surface.Longer SSBs undergo further expansion during epitaxial growth, resulting in increased length with temperature.However, this increment eventually reaches a maximum, likely due to the existence of an upper limit for the etching rate of the 4H-SiC surface within the confines of our experimental parameters.At higher temperatures, the increased vapor pressure of etched species above the surface counteracts the etching process.
On the other hand, an increase in epitaxial layer thickness yields a linear extension of SSB length, indicating the complete replication and expansion of SSBs throughout the growth.High kinetic energy and mobility of growth species can also explain the exceedingly smooth surface and the prevalence of TSD-related surface pits in samples grown using CH 4 as the carbon source.A higher surface mobility may lead to the incorporation of these species into step structures already formed during surface preparation, no new step bunching is formed during growth, and a very low density of step bunching observed in these samples was probably formed during surface preparation which becomes more prominent during growth.The enhanced mobility of molecular species increases the likelihood of a pure step-flow growth mode and an atomically flat surface with relatively large terraces and an increased step height (Figure 11c).Furthermore, a combination of the low sticking coefficient of molecular species decomposed from CH 4 and their high diffusivity together with higher surface energy around the core of TSDs may decrease the vapor pressure of these species and consequently lower growth rates in those regions which may eventually lead to the formation of TSD-related surface pits after the growth.
A similar phenomenon can also explain the dominance of overgrown triangular defects.Figure 13 illustrates the evolution of a triangular defect resulting from a particle (downfall) during epitaxial growth with CH 4 as the carbon source.Initially, a particle on the surface causes the pinning of steps around it during step-flow growth and leads to the formation of a wide terrace with (0001) facet at the down step direction (Figure 13a).With further progress in growth, foreign polytype inclusion, such as 3C, may occur and nucleate through 2D growth on this facet.The molecular species contributing to growth, possessing high surface mobility and encountering step edges and kinks in different directions, lead to growth proceeding along the step-flow direction (Figure 13b).Consequently, the downfall is incorporated into the step-flow growth along all mentioned directions (Figure 13c,d).Finally, the steps that were separated by the particle recombined and continued to grow, resulting in the overgrowth of the nucleated 3C structure (Figure 13e).However, this should be noted that the nucleated 3C continues to grow to the surface over the basal plane on which it was nucleated.

C 3 H 8 as the Carbon Source
In contrast to the epitaxial layers grown with CH 4 , the use of C 3 H 8 as the carbon source resulted in a relatively rough surface with RMS values exceeding 0.8 nm.Additionally, the surfaces exhibited pure step-bunching, and for C/Si ratios above 1.0 ILL defects became dominant.The formation of the step-bunched surface can be explained by considering the low diffusion coefficient and mobility of the molecular species on the surface.With the 4°off-cut substrates, step edges are the dominant sites for growth, although kinks in other directions may also be present.Due to the larger molecular species resulting from the decomposition of C 3 H 8 , [35] their mobility is lower, and they do not possess high enough energy to overcome the step edges higher than a specific unit height.Consequently, they incorporate at the nearest step edge with such a height.This  The absence of SSBs on the surface of samples grown with C 3 H 8 can be explained by considering the higher surface roughness in these samples, which causes the SSBs to be invisible on rough surfaces.
The origin and formation mechanism of ILL surface morphological defects on 4 H-SiC epitaxial layers are discussed in detail in ref. [25].There we have demonstrated that TSDs present in the substrate are responsible for the formation of such defects.Figure 8c shows a TSD-related KOH etch pit adjacent to each ILL defect on the surface.Expanding on the previous explanations regarding the mobility of molecular species, it may also be explained why such defects are only observed on the surface of C 3 H 8 -grown epitaxial layers and are absent in epitaxial layers grown with CH 4 .The spiral around the core of TSD produces a local strain field which affects the surface energy and influences the flow of steps related to the off-cut, as an image force.The direction of image force is defined by the sign of Burger's vector of associated dislocation. [5]When the image force and off-cut-related steps are aligned in the opposite direction, an ILL defect is formed on the surface and when they are in the same direction, no such defect is observed on the surface.Lighter carbon molecular species from CH 4 easily overcome the high surface energy around the core of dislocation and diffuse through the surface, whereas heavier carbon species from C 3 H 8 incorporate locally and lead to the formation of ILL defects.As it was observed in Figure 4a, increasing the C/Si ratio results in an increase in the vapor pressure of heavier species and consequently expansion in the inclined line's length.Furthermore, where this expansion does not show a clear relation to the growth temperature, it increases exponentially with the thickness of the epitaxial layer, likely due to a weaker strain field far from the core of the TSD.
The prevalence of pure 3C-type triangular defects on samples grown with C 3 H 8 can be explained by a similar behavior of the molecular species.Figure 15 illustrates that, resulting from a downfall particle, 3C nucleation occurs on the formed terrace.However, in the case of samples grown with C 3 H 8 , the molecular species do not possess enough mobility to escape this terrace.As a result, the 3C region expands, and this expansion continues until the end of growth.
To comprehensively elucidate the key distinctions between CH 4 and C 3 H 8 in 4H-SiC epitaxy, characteristics of epitaxial layers grown through respective hydrocarbons are summarized in Table 4.

Conclusion
In conclusion, our study has presented essential insights into the different impacts of distinct carbon sources on the growth of 4H-SiC epitaxial layers.It was shown that SSBs are already formed on the surface during surface preparation.The introduction of methane resulted in an extension of pre-existing SSBs, whereas propane effectively eliminated these defects.Conversely, propane yielded ILL surface morphological defects which were absent in the methane case.In terms of surface morphology, methane revealed its efficiency in yielding an exceptionally smooth surface decorated with wider terraces.However, the step edges were slightly higher compared to those formed by propane.Moreover, while propane led to a rough and step-bunched surface, the introduction of methane resulted in the emergence of dislocation-related etch pits on the epitaxial layer's surface.The observed differences in using methane or propane as carbon sources are likely attributed to varying sticking coefficients and surface diffusivity of molecular by-products resulting from the distinct thermal decomposition of these hydrocarbons.Overall, while achieving a smooth and defect-free thick 4H-SiC epitaxial layer presents challenges, our investigation underscores the feasibility of modulating the surface structure by carefully selecting the appropriate carbon source.Through such discerning choices, the tailored surface characteristics of these layers can be aligned with specific applications, offering promising avenues for the future.

Figure 1 .
Figure 1.DF and AFM images of different substrates after surface preparation under a) hydrogen (sample EH), b) hydrogen þ CH 4 (sample EM), and c) hydrogen þ C 3 H 8 (sample EP) ambient.

Figure 2 .
Figure 2. a-e) DIC and AFM images of the sample grown using CH 4 as the carbon source at different C/Si ratios.

Figure 3 .
Figure 3. a) DIC image of a typical triangular defect observed on the samples grown with CH 4 as the carbon source, b) Raman shift spectra obtained from the corresponding regions in (a), and c,d) trans and DF images of the same triangular defect.

Figure 4 .
Figure 4. Variation of SSB length in the samples grown with CH 4 as the carbon source versus different a) C/Si ratios, b) growth temperatures, and c) epitaxial layer's thicknesses.
Figure 8a displays the optical image of the bare substrate (sample EH), etched in molten KOH.It can be

Figure 5 .
Figure 5. a-e) DIC and AFM images of the sample grown using C 3 H 8 as the carbon source at different C/Si ratios.

Figure 7 .
Figure 7. Variation in the length of ILL defects for the samples grown with C 3 H 8 as the carbon source versus different a) C/Si ratios, b) growth temperatures, and c) epitaxial layer's thicknesses.

Figure 8 .
Figure 8. Optical images of different samples after being etched in molten KOH: a) bare substrate surface prepared under H 2 ambient, b) epitaxial layer grown with CH 4 as the carbon source, and c) epitaxial layer with ILL defects, grown with C 3 H 8 as the carbon source.

Figure 6 .
Figure 6.a) DIC image of a typical triangular defect observed in the samples grown with C 3 H 8 as the carbon source, b) Raman shift spectra obtained from corresponding points numbered in (a), c) transmission image, and d) DF image of the same triangular defect.An arrow and dotted line in (d) show an example of discontinued surface step bunching over the triangular defect.

Figure 9 .
Figure 9. AFM images and line profile scans of triangular ridges on in situ surface prepared samples in H 2 ambient, showing a) triangular ridge without a bump and b) triangular ridge with a bump at the apex.

Figure 10 .
Figure 10.Schematic illustration of the different stages a-e) during in situ surface preparation around a typical obstacle such as an embedded foreign particle near the surface of 4H-SiC substrate.

Figure 11 .
Figure 11.AFM images together with line profile scans around a) the center position of an SSB, b) one of the tail regions of the SSB, and c) a narrow cluster of step bunching.

Figure 12 .
Figure 12.Schematic illustration of the conversion of a) a triangular ridge-shaped SSB at the beginning of the growth into b) a deep terrace SSB after epitaxial growth.

Figure 13 .
Figure 13.Visual representation of a) steps pinning around a downfall particle, b) nucleation of foreign polytype (here 3C), and c) overgrowth of the 3C by streamlining the downfall particle with 4H step-flow growth when CH 4 is used as the carbon source.d,e) Continuation of the expansion and complete overgrowth.Red dotted lines represent the extent of the 3C structure expansion on the basal plane and underneath the overgrown 4H.
leads to the formation of a step-bunched surface.Comparing the line profiles of the AFM images shown in Figure11cand 14 confirms that the critical step height for C 3 H 8 -grown samples is lower than that for CH 4 , and the use of C 3 H 8 also results in narrower terraces compared to CH 4 samples.

Figure 14 .
Figure 14.AFM image and line profile scan of the surface of a typical sample grown with C 3 H 8 as the carbon source.

Figure 15 .
Figure 15.Schematic depiction of a) steps pinning around a downfall particle, b) nucleation of foreign polytype (here 3C), and c) expansion of the 3C during epitaxial growth when C 3 H 8 is used as the carbon source.

Table 2 .
In situ surface prepared samples information.

Table 3 .
Sticking coefficients of the main species decomposed from the hydrocarbons.

Table 4 .
Key distinctions between CH 4 and C 3 H 8 in 4H-SiC epitaxy.